Tag Archive for: TOF

If you asked most B2B marketing leaders which channel converts best, they’d say email, or paid search, or word of mouth. Almost none would say “AI search”, and that’s not because AI search isn’t converting. It’s because most GA4 setups can’t see it. TLDR: If you are in B2B marketing read the research cited below.

The numbers are hard to ignore

The most cited data point right now (July 2026) is from the Opollo 2026 AI Search Benchmark Report, which pulled GA4 referral data and CRM attribution from 312 B2B technology firms across North America, Australia and the UK. Visitors sourced from AI platforms converted to qualified leads at 14.2%, against 2.8% for standard Google organic traffic. That’s a 5x gap, and it held even among firms getting 100+ AI-referred sessions a month, which rules out small-sampl size noise.

Ahrefs published a steeper version of the same story from its own funnel: AI referrals made up just 0.5% of total sessions but drove 12.1% of all product signups, a 23x conversion differential. Of course, that’s a single company’s case study in a high-consideration SaaS category, not a cross-industry benchmark. I would treat it as a ceiling, not an average.

Platform-level data from Seer Interactive shows the spread isn’t uniform across AI search results. ChatGPT referrals convert around 15.9%, Perplexity around 10.5%, Claude around 5.0%, Gemini around 3.0%. Where your buyers are asking questions matters. [See image below.]

Compared with the rest of the B2B channel mix, AI referral performance is hard to argue with. Traditional B2B organic search sits around 2.4–2.9%. Paid search runs 1.2–1.5% (up to 5.1% for SaaS at the high end). Paid social averages 2.9%. Display is bottom of the table at 0.3%. Only email at 16.9–19.3% is in the same league as AI referral, and email has the advantage of an already opted-in, brand-familiar list. AI search is generating that conversion rate from net-new discovery. White space traffic from net-new logos is highly valuable.

Down-funnel, one case study (BeRelevant/MarketingStack) found AI-sourced B2B leads progressing from MQL to SQL at 61%, against 24% for SEO-sourced leads, and closing at 42% against 28%. That’s a genuinely striking gap but again, it’s one case study, and the caveat across all of this data is that the evidence for AI-referred leads producing better closed-won revenue at scale is still thin and heavily reliant on isolated vendor reporting. Directionally strong. Not yet fully proven across all B2B brands.

CiteCompass AEO results by AI platform

My hypothesis on why the gap exists: journey compression

The mechanism behind all of this is fairly intuitive. A buyer using Google runs a search, opens six tabs, and does the comparison work themselves, click by click. A buyer using an AI engine gets the comparison done for them inside the chat window: the AI has already read the specs, weighed the options and narrowed the shortlist before a single link gets clicked. By the time that buyer lands on your site, over half of AI-referred B2B sessions skip straight past your homepage and blog content entirely and go straight to pricing or solution pages aligned with the buyer journey stage which that prospect is at. They didn’t need the awareness content. The AI filtered past the awareness content.

That’s genuinely good news for anyone selling considered B2B services using content marketing. It’s also exactly why so many of us aren’t seeing it.

CiteCompass CBJ results by stage

The Digital Attic problem

Here’s the catch, and it’s the part that worries B2B marketers more than conversion stats: GA4 doesn’t have a native “AI search” channel. Out of the box, a visit from ChatGPT, Perplexity or Gemini gets bucketed into Direct, Referral, or Unassigned, depending on whether the AI tool bothered to pass a referrer string at all (many don’t, or didn’t until recently). ChatGPT only started reliably appending source tags to desktop links from mid-2025, and app-based clicks still often carry no referrer data.

Which means your highest-converting channel is very possibly sitting right now inside your “Direct” traffic bucket, quietly outperforming everything else in your funnel, completely unlabelled. This is the Digital Attic problem I keep coming back to: it’s not that the value isn’t there. It’s that it’s stored somewhere you never see.

If you’ve never built a custom channel grouping in GA4 for AI referral sources, there’s a real chance you’re making budget decisions on a dashboard that’s structurally blind to the thing that’s working best.

What this means for being cited, not just ranked

The other implication follows straight from “the AI does the shortlisting before the click”: if you’re not part of what the AI reads and cites when it does that shortlisting, none of this conversion data applies to you at all, because you never make the list. Most AI search results cite six links. So you need to be hitting around 14% of the total results in your niche in order to dominate.

That’s a big content marketing job. It may be editing / reposting – but you have to cover all the bases across the full customer buying journey in order to rank in AI search.

Ranking well on Google no longer guarantees you’re in the pool an AI engine draws from when a buyer asks it to compare vendors. Being found and being cited are becoming two different games, and the second one runs on different rules: clear entity naming, structured comparison content, third-party validation (review platforms, comparison articles, partner directories), and content genuinely built to answer the layered questions buyers are now asking an AI instead of a search box.

Two things worth doing this quarter

Fix the measurement blind spot. Build a custom GA4 channel group that captures known AI referral sources (chatgpt.com, perplexity.ai, gemini.google.com, and so on) so you can actually see this traffic rather than inferring it from an unexplained bump in Direct. You cannot make a channel investment case for something your own dashboard insists doesn’t exist.

Audit what an AI engine would actually find and cite if a buyer asked it to compare you against your competitors right now. Not what ranks. What gets read, understood and repeated. Those are no longer the same test. And yes, I use CiteCompass.com for this work – it costs US$10 to do a one-off scan of your site.

The conversion numbers above are the exciting part of this story. The GA4 gap is the part that decides whether you ever get to benefit from them.